MySQL performance

MySQL is one of the most used open source database systems. Good performances with this software will be beneficial not only to applications that use it directly but, as it is often used as a backend web storage, also to dynamic web page servicing.

We run various test against Red Hat and Debian MySQL installation. The first one is the time needed to populate a test database with 100K rows:

Red Hat vs Debian - Sysbench mysql prepare

As you can see, in this case I distinguished between EXT3 and EXT4 installations because the database tests are a disk/filesystems intensive ones. It seems that, while with the aging EXT3 filesystem Red Hat has some advantage, with the much more modern EXT4 filesystem we have a tie.

Lets see Sysbench simple test (select) results:

Red Hat vs Debian - Sysbench mysql simple

Red Hat seems to have a respectable ~15% advantage here.

What about the complex, transactional Sysbench test?

Red Hat vs Debian - Sysbench mysql complex

This disk-bound test show us no real differences.

Sysbench's results seems to suggest mostly similar performances between the two Linux systems, with some advantages for Red Hat in select-intensive database.

Will the MySQL integrated benchmark suite confirm this?

Red Hat vs Debian - MySQL Bench

It seems not: while the EXT3 Debian system is the slowest, the EXT4 Debian system is faster than the Red Hat one by about (you know!) ~15%.

In the end, it seems that Red Hat and Debian have quite comparable MySQL performances: one workload can favor one of them over the other, but we are not speaking about streaking differences.